Screening for autism spectrum disorders in a developmental clinic in the Western Cape : using the modified checklist for autism in Toddlers (M-CHAT)

ASD has an estimated prevalence of 1 in 68, making it one of the most common neurodevelopmental disorders in children. Furthermore, its prevalence is increasing; therefore, there is a rising demand for screening tools to help achieve beneficial early diagnosis and intervention outcomes. However, there is a lack of literature around ASD and ASD screening tool validity in South Africa. This thesis adapted and assessed the use of South African English, Afrikaans and IsiXhosa versions of the 23-item Modified Checklist for Autism in Toddlers (M-CHAT) screening tool, for a Western Cape state hospital. The M-CHAT was completed by parents (N=255) of children between the ages of 1.5 and 4.99 years, at the Red Cross Children's Hospital developmental clinic. The demographic variables of Child's Age or Sex, Income and Mothers Education did not significantly affect the M-CHAT scores. Furthermore, on qualitative inspection, neither Home language nor Ethnicity of the child appeared to affect the screening scores. Final M-CHAT scores and high internal consistencies were similar across the three M-CHAT language versions, likely indicating their equivalence in flagging ASD. Even the extended age range (4.01-4.99 years) included in this study, did not appear to affect the M-CHAT scores. The M-CHAT follow-up interview was important in determining the ASD risk outcome. Overall, 67% failed the M-CHAT initially, thus requiring follow-up questioning, and of those, 40% changed their outcome and subsequently passed. Interestingly, filling out the M-CHAT in a first or second language did not affect the percentage requiring the follow-up nor the proportions changing their outcome after follow-up. The items which were poor or good discriminators between those eventually passing or failing overall were investigated. The good discriminating items were not necessarily the same as the originally suggested critical items. Thus, new critical items and the possible removal of unnecessary items for this context may need further investigation. For phase 2, a small subgroup (n=38) filling in the English M-CHAT took part in formal ASD diagnostic assessments, using the Autism Diagnostic Observation Schedule (ADOS/ADOS-2). Preliminary investigations into the English M-CHAT's predictive abilities are promising. A cut-off of 3/23 items overall indicates high sensitivity (.84) and adequate specificity (.69). The adapted cut-off of 1/6 critical items results in good sensitivity (.76) and high specificity (.92). The promising results warrant further investigations into the predictive validity of all 3 language versions of the M-CHAT. This thesis takes the first steps in validating the use of the M-CHAT in this low SES context and indicates positive prospects for its future use in state clinics in the Western Cape, and ultimately South Africa